Vol.:(0123456789) 1 3 Metabolomics (2017) 13:146 DOI 10.1007/s11306-017-1286-8 ORIGINAL ARTICLE Automated metabolite identification from biological fluid 1 H NMR spectra Arianna Filntisi 1  · Charalambos Fotakis 2  · Pantelis Asvestas 3  · George K. Matsopoulos 1  · Panagiotis Zoumpoulakis 2  · Dionisis Cavouras 3   Received: 5 June 2017 / Accepted: 19 October 2017 © Springer Science+Business Media, LLC 2017 tools, exhibiting good performance in terms of correct assignment of the metabolites. Conclusions This new robust scheme accomplishes to automatically identify peak resonances in 1 H-NMR spectra with high accuracy and less human intervention with a wide range of applications in metabolic profiling. Keywords 1 H NMR spectroscopy · Metabolomics · Automated metabolite identification · Human Metabolome Database · Spectra preprocessing 1 Introduction Nuclear Magnetic Resonance (NMR) spectroscopy and Mass Spectrometry (MS) have emerged as key technologies for metabolite analysis (Lenz and Wilson 2007; Lindon and Nicholson 2008; Larive et al. 2015) by examining various biofluids and elucidating biomarkers of disease (Fischer et al. 2014; Jobard et al. 2014; Smolinska et al. 2012; Deng et al. 2016; Kang et al. 2015; Kordalewska and Markusze- wski 2015; Psychogios et al. 2011). One of the main issues in these studies is metabolite identification, since interpret- ing 1 H NMR spectra is a challenging, time-consuming task (Li et al. 2013; Smolinska et al. 2012). To this end, computational metabolite recognition has been the objective of numerous research efforts (Domingo- Almenara et al. 2016; Chignola et al. 2011; Mihaleva et al. 2009; http://www.chenomx.com). BQuant is based on Bayesian modelling and addresses metabolite 1 H-NMR detection as a variable selection problem (Zheng et al. 2011). A probabilistic method based on Markov chain Monte Carlo (MCMC) and Metropolis–Hastings block updates has been implemented in the BATMAN package (Hao et al. 2012). Mercier et al. (2011) have proposed an automated spectral Abstract Introduction Metabolite identification in biological sam- ples using Nuclear Magnetic Resonance (NMR) spectra is a challenging task due to the complexity of the biological matrices. Objectives This paper introduces a new, automated compu- tational scheme for the identification of metabolites in 1D 1 H NMR spectra based on the Human Metabolome Database. Methods The methodological scheme comprises of the sequential application of preprocessing, data reduction, metabolite screening and combination selection. Results The proposed scheme has been tested on the 1D 1 H NMR spectra of: (a) an amino acid mixture, (b) a serum sample spiked with the amino acid mixture, (c) 20 blood serum, (d) 20 human amniotic fluid samples, (e) 160 serum samples from publicly available database. The methodo- logical scheme was compared against widely used software Binary file freely available for download at http://biomig.ntua.gr/ downloads/software/MIDTool.zip. Electronic supplementary material The online version of this article (doi:10.1007/s11306-017-1286-8) contains supplementary material, which is available to authorized users. * Panagiotis Zoumpoulakis pzoump@eie.gr 1 School of Electrical and Computer Engineering, National Technical University of Athens, 9 Iroon Polytechniou str., 15780 Athens, Greece 2 Institute of Biology, Medicinal Chemistry and Biotechnology, National Hellenic Research Foundation, 48 Vas. Constantinou Ave., 11635 Athens, Greece 3 Department of Biomedical Engineering, Technological Educational Institute of Athens, 17 Ag. Spyridonos Street, 12243 Athens, Greece